It would be great to see this updated to include LLaMA (and possibly be able to hot-swap other open source models)
Side note: wanted to re-tweak this such that it can mash-up Inspirobot quotes with regular quotes to have more sass.
Question: how would instruction prompting work? Currently I can only think of 4 prompts for authorship styles
Generate a quote in the writing style of X
Generate a quote in the style of X
Generate a quote by X
Generate a X quote
Side note: what is the performance difference between LoRA and direct finetuning?